Research Of Video Traffic Flow Parameters Detection Algorithms

The correct detection and extraction of traffic flow parameters are the important basis for the monitoring and management of highways. The methods of traffic flow parameters detection are diverse, such as the electromagnetic induction coil, ultrasonic detectors, microwave detectors, infrared detectors, as well as the video traffic flow parameters detection, which has been developed most rapidly. Because of the strong points of large detection region, flexible system setting and not affecting the traffic mode, the video traffic flow parameters detection has become a hotspot of research in the traffic fields, and will become the new mode of traffic management in the future.Based on the digital image processing and the computer vision technology, the paper has conducted in-depth research in the key technologies of the system, and has proposed new discussion and methods of key algorithms, such as the background model, the object segmentation, the shadow removal, and the target tracking. In the aspect of system designing, considered the existing monitoring equipments in current transportation fields, the paper has designed the compatibly mode of "the video image gathering module + video detection server + video detection terminal". Besides, examination mechanism, which has based on ROI (Region of Interest) and the ROD (Region of Detection) has been proposed.The research fruits of this paper are as follows. In the study of key algorithms, this paper has improved the object segmentation algorithm by proposing the algorithm based on the average-time method and the statistical background model. In the target tracking aspect, movement target tracking algorithm based on the regional characteristics and Kalman filter has been proposed. In the aspect of the region character, the matching algorithm based on the center distance of mass and RGB color histogram has been proposed. In the shadow removal aspect, the current HSV model based on gradient operator has been researched and optimized. In the system developing, the video detection server and the video detection terminal were developed. Besides, the track algorithm and the system configuration module were developed independently.The road-tests of traffic flow parameters using the algorithms of this paper have been done on the highway. The results indicate that traffic flow parameters have higher precision and can be further used in practical engineering.In addition, this paper has conducted deep analysis on the factors which would affect the system performance. Besides, the optimal angle of the camera installation has been proposed.